CRUST 5.1: A global crustal model at 5 ø x 5 ø

We present a new global model for the Earth's crust based on seismic refraction data published in the period 1948-1995 and a detailed compilation of ice and sediment thickness. An extensive compilation of seismic refraction measurements has been used to determine the crustal structure on continents and their margins. Oceanic crust is modeled with both a standard model for normal oceanic crust, and variants for nonstandard regions, such as oceanic plateaus. Our model (CRUST 5.1) consists of 2592 5 ø x 5 ø tiles in which the crust and uppermost mantle are described by eight layers: (1) ice, (2) water, (3) soft sediments, (4) hard sediments, (5) crystalline upper, (6) middle, (7) lower crust, and (8) uppermost mantle. Topography and bathymetry are adopted from a standard database (ETOPO-5). Compressional wave velocity in each layer is based on field measurements, and shear wave velocity and density are estimated using recently published empirical Vp-V s and Vp-density relationships. The crustal model differs from previous models in that (1) the thickness and seismic/density structure of sedimentary basins is accounted for more completely, (2) the velocity structure of unmeasured regions is estimated using statistical averages that are based on a significantly larger database of crustal structure, (3) the compressional wave, shear wave, and density structure have been explicitly specified using newly available constraints from field and laboratory studies. Thus this global crustal model is based on substantially more data than previous models and differs from them in many important respects. A new map of the thickness of the Earth's crust is presented, and we illustrate the application of this model by using it to provide the crustal correction for surface wave phase velocity maps. Love waves at 40 s are dominantly sensitive to crustal structure, and there is a very close correspondence between observed phase velocities at this period and those predicted by CRUST 5.1. We find that the application of crustal corrections to long-period (167 s) Rayleigh waves significantly increases the variance in the phase velocity maps and strengthens the upper mantle velocity anomalies beneath stable continental regions. A simple calculation of crustal isostacy indicates significant lateral variations in upper mantle density. The model CRUST 5.! provides a complete description of the physical properties of the Earth's crust at a scale of 5 ø x 5 ø and can be used for a wide range of seismological and nonseismological problems. 1. Introduction There are numerous applications for a global model of the seismic velocity and density structure of the Earth's crest and uppermost mantle. In the field of seismology, such a model provides regional travel times for locating earthquakes. Obviously , a good crustal model is the key ingredient to successfully monitoring regional-scale seismicity. Most mantle seismic to-mographic methods use data sets which are quite sensitive to crustal structure but, at the same time, cannot resolve details within the crest. Hence accurate "crustal corrections" applied to these data sets are essential to improving the resolution of even large-scale mantle structure. Lateral variations in mantle density may also be inferred from long-wavelength gravity data if the density structure and thickness of the crust are reasonably well known. The crustal contribution to lithospheric stress and crustal isostasy can be calculated from crustal thickness, density, and topography. Previous global crustal models have provided various levels of detail. Sollet et al. [1982] presented a crustal thickness map but did not specify seismic velocities or densities. Hahn et al. [1984] presented a model wherein the crustal structure was described in terms of irregularly shaped regions, each with a uniform structure. More recently, Tanimoto [1995] reviewed the crustal structure of the Earth using a wide range of seismic data, and Nataf and Ricard [1996] presented a model for the crust and upper mantle on a 2 ø x 2 ø scale (3SMAC). This latter model was derived using both seismological data and nonseis-mological constraints such as chemical composition, heat flow, and hotspot distribution, from which estimates of seismic velocities and the density in each layer were made. In this paper, we present a new global crustal model (CRUST 5.1) that is based on significantly more data than previous models. Compiling a new global crustal model is timely because of the availability of a large body of new data. 727